2018#

View page source

Release Notes for 2018#

mm-dd#

11-10#

  1. Added the dynamic argument to the independent function and the f_new_dynamic function.

  2. Added the sections more.cpp and more.py. These have been replaced by new_feature .

11-09#

Include binary operations where first operand is an a_double and second operand is a double ; see a_double_binary, a_double_compare, and a_double_assign.

11-07#

  1. Advance to cppad-20181106 (must re-run bin/get_cppad.sh ).

  2. The a_fun objects were added. This support all the operations that d_fun support all the py_fun operations with the exception of optimize. The py_sparse operations are not yet supported.

11-05#

  1. API Change : change cppad_py.a_fun to cppad_py.d_fun because it is a function that evaluates in double and plan to use cppad_py.a_fun for a function that evaluates in a_double .

  2. Advance to cppad-20181105 (must re-run bin/get_cppad.sh ).

08-13#

  1. Change cppad_cxx_flags to extra_cxx_flags in the user configuration settings. Also remove swig_cxx_flags from these user settings (set automatically).

  2. Improve the install instructions; see old_setup.py.

07-31#

  1. The general purpose Swig example swig_xam , which was not specific to cppad_py, has been removed.

  2. A change to old_setup.py was making the get_cppad.sh script fail. This has been fixed.

  3. The python tests are no run in example/python (not copied to the build directory).

07-26#

Change cppad_py/python_major_version to cppad_py/python_version because it now contains both major and minor version numbers.

07-19#

Add -std=c++11 to the default compile flags in old_setup.py. Add -py3 to the swig builds when running old_setup.py using Python 3; see swig python 3.

07-15#

  1. Better error reporting for type and size errors in numpy.array arguments to cppad_py functions using numpy2vec.

  2. Finish converting all vectors and matrices in py_lib to numpy arrays. Note the py_utility routines do the conversion.

07-14#

  1. Continue conversion of python library to using numpy.array ; see py_fun_forward, py_fun_reverse.

  2. Add the f_size_order() function.

07-13#

Continue conversion of python library to using numpy.array ; see py_fun_jacobian, py_fun_hessian.

07-12#

  1. Continue conversion of python library to using numpy.array ; see py_independent, py_fun_ctor.

  2. Add the file cppad_py/python_major_version which contains the major version of python that this build of cppad_py is for.

  3. Automatically use build_type control how old_setup.py compiles C++ (one used to use flags on old_setup.py command line).

  4. old_setup.py was using the git repository version of CppAD instead of the one installed in build/prefix by get_cppad.sh. This has been fixed.

  5. Change size_ind to size_domain and size_dep to size_range.

07-10#

Start conversion of python library from using vector to using numpy.array ; see py_independent.

07-08#

  1. Advanced to cppad-20180703 .

  2. Fix the following warning: catching polymorphic type class 'std::runtime_error' by value .

  3. Suppress the following warning when running old_setup.py: clearing an object of non-trivial type class 'cppad_py::a_double' ; see swig issue 1259.

  4. Move all settings to old_setup.py and run cmake from there.

  5. The file cppad_py/__init__.py was not being created. This has been fixed.

07-07#

Change build of Python module to use Python 3 (now it can use either 2 or 3).

07-03#

Birthday when CppAD Py was first extracted from CppAD Swig.